Integrated targeted serum metabolomic profile and its association with gender, age, disease severity, and pattern identification in acne

Background Westernized diet and nutritional metabolism are important in acne pathogenesis, especially in adult patients. However, clinical and basic data are lacking. Pattern identification (PI) is a tool that results in a diagnostic conclusion based on a cluster of concurrent symptoms and signs in traditional medicine. Acne can be classified by PI. However, whether the metabolomic profile differs according to the PI of acne is unknown. Metabolomic data would help clarify the pathogenesis of acne. Methods We conducted a cross-sectional study involving 40 healthy controls and 60 subjects with acne. We evaluated androgens, serum lipids, essential amino acids, nonessential amino acids, other amino acids, and pro-inflammatory cytokines of all subjects and compared the metabolomic profiles between acne subjects and healthy controls, and in subgroups according to gender, age, severity, and PI. Results Dehydroepiandrosterone sulfate and serum fatty acids were significantly higher in female subjects, adolescents, and those with disharmony of the thoroughfare and conception vessels. The total essential and nonessential amino acids were significantly lower in the overall, female, adult, severe, and phlegm-stasis group. The latter group exhibited elevated serum levels of interleukin-1β and -6. Conclusions This is the first study to investigate serum lipids, amino acids, and cytokines in subjects with acne. We analyzed the differences between metabolomic profiles to determine the diagnostic value of PI. Prospective studies with more patients are needed to confirm the characteristics of each PI and lipidomic data will enrich knowledge concerning lipid mechanism.


Introduction
Acne vulgaris is a common chronic inflammatory dermatosis that affects 85% of adolescents and young adults worldwide [1][2][3]. Increased risk of acne has been described in developed countries compared with developing countries [4], which is closely related with the adoption of a more westernized diet. A recent study reported that acne prevalence in adults is increasing, especially in female subjects [5]. Adult acne or adult female acne is considered as a specific acne subtype distinct from adolescent acne [6]. Diet and nutritional metabolism play key roles in acne pathogenesis, especially in adult patients. However, clinical and basic research data are lacking.
In addition to diet, hormonal abnormalities [7], stress and modern life style [8], smoking [9], environmental factors [10,11], and genetic predispositions [12,13] may contribute to acne. The influence of these systemic factors on acne has recently changed the perception of acne as both a site-specific and systemic disease [14]. From this perspective, the development of a tool that classifies acne according to systemic symptoms is needed for the individualized and targeted therapy of acne. Pattern identification (PI) is a tool that results in a diagnostic conclusion based on a cluster of concurrent symptoms and signs in traditional medicine that is based on the themes of holism and individualization [15]. Although standardized diagnostic criteria have not been described in the classic literature, advancements in systems biology and computing techniques have led to progress in the standardization and modernization of PI [16,17]. A standardization study revealed that acne also can be classified by PI [18]. However, the PI metabolomic profile indicating the difference of pro-inflammatory cytokines in acne is unknown.
Metabolomic analysis would provide more knowledge of the pathogenesis of acne. Thus, the aims of this study were to compare the metabolomic profiles with pro-inflammatory cytokines between acne subjects and healthy controls, and to compare the differences in these metabolomic profiles with pro-inflammatory cytokines in subgroups according to gender, age, severity, and PI.

Study design
This cross-sectional study was conducted at Kyung Hee University Korean Medicine Hospital (Clinical Research Information Service, Republic of Korea: KCT0002212) [19]. Healthy volunteers and subjects with acne were enrolled at a screening visit and outcomes were measured after one week (Table 1). This study was approved by the institutional review board of Kyung Hee University Korean Medicine Hospital (KOMCIRB-161118-HR-062). Written informed consent was obtained by the investigator from all participants prior to enrollment. of healthy controls. Moreover, in the metabolomic study design, a sample size of three to five subjects per group may provide useful preliminary data [20]. Therefore, in this exploratory study, the number of subjects was set as 40 healthy subjects and 60 acne subjects, which enables the assessment of the differences in biomarkers between healthy individuals and acne subjects, including a respective 95% confidence interval, with sufficient precision according to the size of the metabolite kit, with expert opinions in metabolomics, molecular biology, dermatology, and statistics. The gender ratio in each group was 1:1.

Subjects
The inclusion criteria for acne group were as follows: (1) age 19-35 years, (2) body mass index (BMI) 18.5 to 23 kg/m 2 , and (3) voluntary participation in this clinical trial. The inclusion criteria for the control group were as follows: (1) age 19-35 years, (2) BMI 18.5 to 23 kg/m 2 , (3) no evidence of acne, (4) (only for female subjects) regular menstrual cycle during the last 3 months, (5) no digestive-related uncomfortable symptoms in usual situations, (6) no abdominal muscle tension, (7) other disorders that may affect the outcomes, and (8) voluntary participation in this clinical trial. The exclusion criteria for both groups were: (1) use of oral steroids, oral contraceptives, oral vitamin A derivatives, antibiotics, or herbal medicines that may affect the test results during the last 4 weeks, (2) no fasting for more than 8 hours after 9:00 pm on the day before the visit, (3) consumption of alcohol in the preceding week, (4) smoking, (5) pregnancy or lactation, (6) previously diagnosed with polycystic ovary syndrome (PCOS), (7) judgment that physical or mental testing is not appropriate for the clinical trial, and (8) participation in other clinical trials that might affect this trial.

Outcome
Metabolomic profiles with pro-inflammatory cytokines. We evaluated two androgens (dehydroepiandrosterone sulfate [DHEA-S] and testosterone), serum lipids, six essential amino acids (EAAs), seven nonessential amino acids (NEAAs), three other amino acids (AAs), and eight pro-inflammatory cytokines in all subjects. More information about profiling methods are presented in a previously published protocol study [19]. Subjects in each group were divided into two groups: adolescent acne (age �25 years) and adult acne (age >25 years) [21].
PI. The assessor fulfilled the inventory for PI of acne, which is based on weighted scoring system (see S1 Table) after diagnosis of each subject [18]. After completing the questionnaire,  Table 2). The DTCV acne group included only female subjects, since the diagnosis is related to the menstrual cycle. Therefore, values of the DTCV acne group were compared with those of female subjects in the control group. Values of female subjects in the other PI groups were also compared with those of female subjects in the control group, to assess whether the characteristics of the female subjects in the DTCV group differed from other female subjects with acne. Disease severity. Acne severity was defined in accordance with the Korean Acne Grading System. The grade ranged from 1 to 6, where 1 = � 10 papules, 2 = 11-30 papules, 3 = � 31 papules; � 10 nodules, 4 = 11-20 ± mild ongoing scars, 5 = 21-30 nodules ± moderate ongoing scars, and 6 = � 31 nodules ± severe ongoing scars ± sinus tract (papule; acne < 5 mm, nodule; acne > 5 mm) [22]. Subjects were divided into two groups: mild acne (grade 1-3) and severe acne (grade 4-6) Skin assessments. Skin assessments for skin temperature, skin stiffness (cutaneous resonance running time), and sebum levels were performed on the both facial cheeks using a model ST500 skin thermometer, model RVM600 reviscometer, and model SM815 sebumeter SM815 (all from Courage + Khazaka Electronic GmbH, Kern, Germany), respectively [23,24]. The assessments were performed three times and average values were used. The assessments were performed under strictly controlled conditions with a room temperature of 22 ± 2˚C and a relative humidity of 50 ± 10% and after the resting period of approximately 30 min.

Statistical analyses
All statistical analyses were performed using SPSS 21 software (IBM Inc., Armonk, NY, USA). Data are presented as means ± standard deviations for continuous data or number (n) for categorical data. The baseline characteristics were compared by an independent t test for continuous values and chi-square test for categorical values. An independent t test was performed for the comparison of continuous data between the healthy subjects and acne subjects and subanalyses according to gender, age, and PI. Analysis of covariance was performed for the comparison according to severity (controls vs. mild vs. severe), after adjustment for gender. Tukey's HSD post-hoc test was used to determine which groups differed from each other. In all tests, a p-value < 0.05 was considered statistically significant.

Subjects
From May 2017 to November 2017, a total of 40 healthy volunteers and 60 of acne subjects were included in this study. There were no significant differences in BMI, gender, smoking,

Wind-Heat
Acne type mainly with papules that are reddish, painful with itching, have burning sensations, and sometimes have pustules.

Dampness-Heat
Acne type mainly with pustules that are easily swelling, red, painful with greasy skin, and sometimes have nodules.
Phlegm-Stasis Acne type mainly with nodules or cysts that are hard and painful, dark in color, can be long lasting, recur easily, and remain as scars, pimples, and pigmentation.

Disharmony of the thoroughfare and conception vessels
Acne type that is closely related with the menstrual cycle, is especially exacerbated before menstruation, and tends to occur in the form of small papules located under the chin or around the mouth.
alcohol consumption, sleep time, stress level, dietary habits, and skin assessment between two groups. A difference was evident for age ( Table 3).

Comparison of metabolomic profiles with pro-inflammatory cytokines between groups, and gender and age subgroup analysis
Overall. No significant differences were observed in the androgens and cytokines. Serum lipids and linoleic acids were significantly higher in the acne group. In the acne group three EAAs, two NEAAs, sum of EAAs, and sum of NEAAs were significantly lower, while two NEAAs and one of the other AAs were significantly higher (Table 4 and Fig 1).
Male subgroup. No significant differences were noted in androgens, serum lipids, EAAs, and cytokines. Two NEAAs and sum of NEAAs were significantly lower in the acne group (Table 4 and Fig 1).
Female subgroup. No significant differences were found in the cytokines. Among androgens, DHEA-S, and among serum lipids, palmitic acid were significantly higher in the acne group. Three EAAs and sum of EAAs were significantly lower, while one NEAA and one of the other AAs were significantly higher in the acne group (Table 4 and Fig 1).
Adolescent subgroup. No significant differences were observed in EAAs and cytokines. Among androgens, DHEA-S, and among serum lipids, linoleic and oleic acids, were significantly higher in the acne group. One NEAA was significantly lower in the acne group (Table 4 and Fig 1).  Adult subgroup. No significant differences were found in androgens, serum lipids, and cytokines. Five EAAs, two NEAA, sum of EAAs, and sum of NEAAs significantly lower, while two NEAAs were significantly higher in the acne group (Table 4 and Fig 1).

Comparison of metabolomic profiles with pro-inflammatory cytokines according to disease severity
No significant differences were found in androgens and cytokines. Among serum lipids, linoleic acids were significantly higher in the mild group than in the control group. One EAA and one NEAA were significantly lower in the mild group than in the control group. One EAA, one NEAA, sum of EAAs, and sum of NEAAs were significantly lower, while two NEAAs and one of the other AAs, were significantly higher in the severe group than in the control group (Table 5 and Fig 2).

Comparison of serum metabolites and pro-inflammatory cytokines according to PI
Wind-heat. No significant differences were noted in androgens, serum lipids, and cytokines. One NEAA and one of the other AAs were significantly higher in the WH acne group than in the control group (Table 5 and Fig 2). Dampness-heat. No significant differences were noted in androgens, serum lipids, and cytokines. Three EAAs and one NEAA were significantly lower in the DH acne group than in the control group (Table 5 and Fig 2).
Phlegm-stasis. Subjects with severe acne symptoms displayed no significant differences in androgens. Among serum lipids, cholesterol was significantly lower in the PS acne group than in the control group. Two EAAs, three NEAA, and sum of NEAAs were significantly lower, while one EAA, three NEAAs, and one of the other AAs were significantly higher in the PS acne group than in the control group. Among cytokines, interleukin (IL)-1β and IL-6 were significantly higher in the PS acne group than in the control group (Table 5 and Fig 2).

Disharmony of the thoroughfare and conception vessels
Among androgens, DHEA-S was significantly higher in the DTCV acne group than in the female subjects of the control group, while no significant differences were observed among the female subjects in other PI groups. Among serum lipids, palmitic acid was significantly higher in the DTCV acne group than in the female subjects of the control group, while no significant differences were observed among the female subjects of other PI groups. Among EAAs, isoleucine, tryptophan, and sum of EAAs were significantly lower in the DTCV acne group than in the female subjects of the control group. Isoleucine, phenylalanine, and sum of EAAs were significantly lower in the female subjects in other PI groups than in those of the control group. Among NEAAs, sarcosine was significantly higher in the female subjects of other PI groups and hydroxyproline was significantly higher in the DTCV acne group than in the female subjects of the control group. Among cytokines, IL-15 was decreased in the DTCV acne group, while IL-1β was increased in the female subjects in the other PI groups compared with the control group (Table 5 and Fig 2).

Discussion
Acne vulgaris has been suggested to be closely related with mammalian target of rapamycin complex 1 (mTORC-1) driven metabolic diseases [25], which include obesity [26], type 2 diabetes mellitus [2], and cancer [27]. Western diet, which is characterized by the high intake of hyperglycemic carbohydrates and dairy products, increases insulin and insulin-like growth factor-1 (IGF-1) [28,29], and promotes mTORC-1 signaling, the key regulator of anabolism and lipogenesis [30,31]. mTORC-1 activation induces hyperseborrhea and hyperkeratinization of the pilosebaceous duct by stimulating androgens, and dysseborrhea by changing the  Integrated targeted serum metabolomic profile in acne proportion of free fatty acids (FFAs) in sebum. IGF-1 induces pro-inflammatory cytokine expression in sebocytes by activating the nuclear factor-kappa B pathway [25].
In this study, DHEA-S was elevated only in female subjects and adolescents with acne. DHEA-S was not elevated in the overall subjects, male subjects, and adults with acne. Previous studies reported that the rise in DHEA-S correlates with the onset of acne in children, and higher levels of DHEA-S are found at puberty in girls with acne than in those without acne [32]. Subjects with the highest levels of DHEA-S develop the most severe acne among adolescent girls [33]. Hormonal factors are also important in adult acne. It was reported that 62% of adult women with acne experience worsening of their acne in the days before menstruation [34]. However, the majority of adult acne subjects have normal androgen levels in the absence of PCOS, congenital adrenal hyperplasia, and adrenal or ovarian tumors [35]. This suggests that initiation of acne symptoms is induced with elevated androgens in puberty, especially in girls, and with the persistent progression of acne into adulthood; hormonal receptors expressed by both sebocytes and keratinocytes may be more sensitive to low levels of androgens [35,36]. Meanwhile, only female subjects in the DTCV group displayed elevated DHEA-S compared with those in the control group, despite the larger proportion of subjects with mild symptoms. The DTCV pattern displayed a weighted value for the effect of the menstrual cycle concerning the worsening of acne, menstrual irregularity, and pain in the lower abdomen (See S1 Table). Among these symptoms, menstrual irregularity is the most representative symptom of PCOS. Although we excluded subjects with previously diagnosed PCOS and obese subjects during enrollment, subjects with undiagnosed PCOS may not have been fully excluded, since PCOS is a common but often undiagnosed condition [37]. Therefore, many female patients with undiagnosed PCOS visit dermatological clinics to treat their acne. Dietary correction and hormonal therapy are more helpful for patients with PCOS. However, hormonal testing is not frequently performed due to its cost and inconvenience of blood collection. Therefore, determining the PI pattern after the completion of a single questionnaire may improve the clinical efficiency of treatment for female patients.
Several studies have reported quantitative changes of sebum (hyperseborrea) and qualitative modifications of sebum (dysseborrhea) with higher amounts of FFA in skin affected by acne [38][39][40]. Elevated IGF-1 and insulin that is mediated by western diet increases sebaceous gland growth and sebaceous lipogenesis, which causes hyper-and dysseborrhea in acne patients [25]. Concerning serum lipids, acne patients frequently display abnormal lipid profiles, including total cholesterol, triglyceride, low density lipoprotein cholesterol, and lipoprotein [41,42]. The present study is the first to evaluate serum FFAs in patients with acne. The data are consistent with the observation of elevated FFAs in insulin resistant subjects [43], and the increases of linoleic acids (polyunsaturated FFA), oleic acids (monounsaturated FFA), and palmitic acids (saturated FFA) in acne subjects. These findings indicate that serum FFAs as well as sebum FFAs are elevated in patients with acne. Consistent with results for DHEA-S, FFAs were elevated only in female subjects and adolescents, not in the male subjects and adults. Androgens stimulate lipid synthesis [44], and the upregulation of androgens and serum lipids affects acne in female subjects and adolescents. Additional lipidomics data are needed to clarify the lipid mechanism in acne subjects.
No studies have reported the profile of serum AAs in acne. In this study, sum of EAAs and NEAAs were decreased in the overall, female, adult, severe acne, and PS acne subgroups. The PS group had weighed value for inflammatory lesion, pustule, cyst, nodule, and pigmentation (see S1 Table). Consistent with our observations, decreases in serum AAs have been described for several chronic inflammatory disorders, including rheumatoid arthritis, inflammatory bowel disease, and chronic kidney disease [45][46][47]. Concerning dermatology, AAs chelate calcium ions and stimulate exfoliation of the stratum corneum, accelerating the turnover of the stratum corneum [48]. Therefore, decreased serum levels of AAs may induce hyperkeratinization, one of the four major factors of acne. However, topical exfoliating therapies can have a negative impact on the barrier functions by increasing transepidermal water loss [49]. Our findings indicate that AAs may be efficiently used to improve skin affected by acne, especially in adults, females, and those with severe acne.
IGF-1 and Cutibacterium acnes are the most important factors that induce an inflammatory response in acne. A recent study of biopsy samples from acne subjects showed that expression of pro-inflammatory cytokines is upregulated after stimulation of IGF-1 [50,51]. C. acnes induces production of local pro-inflammatory cytokines, including IL-1β, IL-8, IL-12 and tumor necrosis factor-alpha (TNF-α) via a Toll-like receptor 2 dependent mechanism [52,53]. The present study is the first to evaluate the serum levels of IL-1α, IL-1β, IL-6, IL-8, IL-12β, IL-15, TNF-α, and granulocyte-macrophage colony-stimulating factor in acne subjects compared with the levels in healthy controls. No significant differences were observed in the acne group and all subgroups, including the severe acne subgroup. However, the PS group, which had weighted values for inflammatory lesions, pustules, cysts, nodules, and pigmentation (see S1 Table), exhibited elevated serum levels of IL-1β and IL-6. This suggests that IGF-1 and C. acnes activate local inflammation in skin lesions. However, the systematic impact may not be significant, except in patients who develop many inflammatory lesions.
There are several limitations to this study. First, 60 acne subjects were divided into four PI groups. The number of subjects in each PI group may not have been large enough to obtain statistical significance. Secondly, acne subjects with undiagnosed PCOS may not have been fully excluded, and this could have biased the analysis of female subjects. Nevertheless, in the real world, blood or ultrasound tests are not frequently performed for patients. Therefore, including subjects with undiagnosed PCOS could better reflect actual clinical practice. This is the first study to investigate serum lipids, AAs, and cytokines in subjects with acne. As an exploratory study, additional dedicated studies are needed to confirm the results. Prospective studies with more patients are needed to confirm the characteristics of each PIs, and future studies that provide more lipidomics data may clarify the lipid mechanism.
Supporting information S1